Recipes / Trigger-event scoring: rank today's accounts by funding, hiring or leadership change

Trigger-event scoring: rank today's accounts by funding, hiring or leadership change

This recipe uses the official PredictLeads, Perplexity and HubSpot MCP servers, and all of them work in Claude. Steps that write, send or spend credits wait for your approval.

Who is this for?

SDRs with a fixed list of target accounts who want to know which ones had a funding round, a hiring push or a new leader worth a call today. PredictLeads can be swapped for Crunchbase or ZoomInfo, and HubSpot for Salesforce, Attio or Pipedrive.

What do you get?

Today's ranked call list (nothing written to HubSpot yet). Sample output: every company and person in it is fictional.

25 target accounts checked. 7 have a signal from the last 30 days. 2 are owned by another rep and are left out.

| # | Account | Score | Signal (date) | Confirmed | Owner / stage |
|---|---|---|---|---|---|
| 1 | Mossgate Labs (fictional) | 6 | Series B announced (Oct 9); 4 new sales ops roles posted | Yes, 2 sources | You / Lead |
| 2 | Larkspur Freightworks (fictional) | 4 | New VP Revenue, Dana Okafor (Oct 2) | Yes, 1 source | You / MQL |
| 3 | Brindle Harbor Foods (fictional) | 2 | 6 open SDR roles | Not checked (jobs only) | You / Lead |
| 4 | Fernhill Software (fictional) | 0 | Funding news from March, outside the window | Old, dropped from score | You / Lead |
...3 more

Why call Mossgate today: new funding and a sales ops hiring push usually mean new tooling budget. Sources: 2 cited articles, linked above.

PredictLeads requests used this run: 75. Perplexity searches: 6.
Reply 'log notes 1, 2' to add the score and sources as a note on those HubSpot companies.

How long does it take, and what does it cost?

Estimate About 15–20 minutes for 25 accounts, most of it checking the shortlist.

Assumptions: One morning run over 25 target accounts, with about 6 shortlisted accounts confirmed and up to 6 notes logged.

• PredictLeads: About one request per signal type per account (jobs, news, financing), so roughly 75 for 25 accounts. Each request counts toward the plan's monthly limit. Every call uses credits. Each MCP request counts toward the plan's monthly request limit. Once the limit is reached, requests return HTTP 402. PredictLeads pricing: 1 API credit per API request; discovery endpoints (e.g. Similar Companies) are billed per returned company. Source: docs.predictleads.com, checked 2026-10-16
• Perplexity: One search or ask call per shortlisted account, about 6 per run. Every call uses credits. Tool calls are billed at standard Perplexity API pricing to the connected API key or to the API organization chosen at sign-in; the organization's rate limits apply. Source: docs.perplexity.ai, checked 2026-10-16
• HubSpot: Reads for 25 companies (owner, stage, open deals), plus up to 6 notes after you approve them. Credit use not confirmed by HubSpot.

Exact credit costs depend on your plan with each vendor. Check the tool pages before a large run.

Which MCP servers does it use?

Every server is confirmed in: Claude

• PredictLeads Not probed
  ClaudeChatGPTClaude CodeAccess not confirmedEvery call uses credits
• Perplexity Up · auth
  ClaudeClaude CodeCursorRead onlyEvery call uses credits
  https://api.perplexity.ai/mcp
• HubSpot Up · auth
  ClaudeChatGPTRead + writeCredits not confirmedNeeds admin or IT
  https://mcp.hubspot.com

How do I set it up in a chat app?

Every server in this recipe is confirmed for Claude. Connect each one, then paste the prompts below into one conversation.

• Connect PredictLeads without code (Claude, ChatGPT)
• Connect Perplexity without code (Claude)
• Connect HubSpot without code (Claude, ChatGPT)

Developer setup

{
  "mcpServers": {
    "predictleads": {
      "url": "None"
    },
    "perplexity": {
      "url": "https://api.perplexity.ai/mcp"
      // headless alternative: "headers": { "Authorization": "Bearer YOUR_PERPLEXITY_API_KEY" }
    },
    "hubspot": {
      "url": "https://mcp.hubspot.com"
    }
  }
}

Step-by-step prompts

Copy all prompts

1. Here are my target accounts: {account list with domains}. Find each one in HubSpot and show me a table with the owner, lifecycle stage and any open deal stage. List accounts owned by someone else separately. Don't change anything.
2. For the accounts owned by me or unowned, check PredictLeads for job openings, news and financing from the last {30} days. First tell me how many PredictLeads requests that will take, and run it only after I approve. Show each signal with its date, the matched domain and a source link.
3. For accounts with a funding or leadership-change signal, use Perplexity to confirm it with a cited source. Tell me how many searches that is and run them only after I approve. Mark anything you can't confirm as 'unconfirmed'.
4. Score each account: confirmed funding 3, confirmed new leader in {department} 2, 3 or more open roles in {department} 1, unconfirmed signals 0. Rank them and give one line on why to call each today, with the sources.
5. Only after I approve it, log a short note on each HubSpot company I pick with the score, the signals and their source links. Show me the note text first.

Can I run it on a schedule? (optional)

• Run it each morning before your first call block, or twice a week if your list changes slowly.
• In a scheduled run, stop at the ranked list and send it to yourself. Log nothing in HubSpot until you approve it.
• Put the account cap in the prompt, for example 25. Every PredictLeads request counts toward the monthly limit, and once it's reached requests return HTTP 402.
• Perplexity bills each call to your API key or API organization. Test one manual run before you schedule it so you know what a run uses.

Where does this go wrong?

• Signals for the wrong company. Similar names (a 'Mossgate Labs' and a 'Mossgate Lab Supply') get mixed up. Give domains, not names, and have it show the domain each signal was matched to. Drop any row where the domain doesn't match.
• Old news scored as new. A funding round from last spring can show up again in a recent article. Require the event date, not the article date, and drop anything outside your window before scoring.
• Hiring noise. Some companies keep the same roles posted all year. Ask for the number of posts and the departments. Only score hiring in the team you sell to, and treat 1–2 posts as no signal.
• Calling into another rep's account. Check the HubSpot owner before anything else. Leave out accounts owned by someone else, or list them separately so you can hand the signal to that rep.

Gotchas

• PredictLeads doesn't publish its MCP endpoint URL. Connect it from the Claude connectors directory or the ChatGPT plugins directory instead. On ChatGPT Enterprise or Edu, an admin may need to enable it first.
• We couldn't confirm whether PredictLeads' MCP is read-only, so this recipe only asks it to read. Discovery lookups like Similar Companies are billed per returned company, so leave them out of a daily run.
• Perplexity's OAuth sign-in needs you to be an admin of a Perplexity API organization that can pay for usage. Otherwise connect with an API key. On claude.ai it's added as a custom connector with its URL.
• HubSpot non-admin users need a Super Admin to approve the connector before they can connect.

Download

Get the prompts, guardrails and expected output as one Markdown file. Paste it into a Claude Project or ChatGPT project's instructions, or save it as CLAUDE.md for Claude Code.

Download trigger-event-scoring.md

Vendor sites: Perplexity · HubSpot. If we join an affiliate program, these links will be labeled as affiliate links.